install Pytorch and CUDA 8.0 with cuDNN 6.0 for Ubuntu 16.04

Install PyTorch

============================================================

Option 1, Install PyTorch via Conda (Do not use sudo)

Install Anaconda

bash conda_pytorch/Anaconda2-4.3.1-Linux-x86_64.sh

Install PyTorch

conda install conda_pytorch/cuda80-1.0-0.tar.bz2
conda install conda_pytorch/pytorch-0.1.12-py27_2cu80.tar.bz2
conda install conda_pytorch/torchvision-0.1.8-py27_2.tar.bz2

Option 2, Install PyTorch via Pip

Update Pip

sudo pip install –upgrade pip -i https://pypi.tuna.tsinghua.edu.cn/simple

Install Scipy Stack

使用清华源镜像加速
sudo pip install -i https://pypi.tuna.tsinghua.edu.cn/simple some-package
some-package: numpy scipy matplotlib ipython==5.1.0 jupyter pandas sympy nose

Install PyTorch

sudo pip install pip_pytorch/torch-0.1.12.post2-cp27-none-linux_x86_64.whl
sudo pip install pip_pytorch/torchvision-0.1.8-py2.py3-none-any.whl

============================================================

Install CUDA 8.0 with cuDNN 6.0 for Ubuntu 16.04

============================================================

PPA

sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt-get update

Driver

sudo ubuntu-drivers devices
sudo apt-get install nvidia-XYZ

CUDA

sudo dpkg -i cuda-repo-ubuntu1604-8-0-local-ga2_8.0.61-1_amd64.deb
sudo apt-get update
sudo apt-get install cuda

cuDNN

sudo dpkg -i libcudnn6_6.0.21-1+cuda8.0_amd64.deb
sudo dpkg -i libcudnn6-dev_6.0.21-1+cuda8.0_amd64.deb

Reboot

sudo reboot

============================================================